Surgical and radiosurgical planning in tumor patients is a very difficult task, requiring the surgeon to mentally form a picture of the complex three-dimensional geometry of anatomical and functional units of the brain, not only around the tumor, but also along all possible paths that access the tumor. Moreover, the surgeon must evaluate each possible surgical strategy in terms of potential complications and probability of a successful outcome. Statistical atlases of the brain can be very helpful in this procedure, as they provide a plethora of useful information, ranging from functional neuroanatomy to statistical models linking tumor parameters, surgical approach, and outcomes. Adapting a statistical atlas of neuroanatomy to a patient's images is a very challenging task, due to the complexity of human brain anatomy and to the confounding effects of edema and extreme deformations in the vicinity of the tumor. This task is undertaken by the proposed project. We propose to build and extensively validate computational models and algorithms that 1) reconstruct a mathematical representation of a tumor patient's anatomy from tomographic images, 2) estimate the deformation that has been induced by tumor growth, 3) adapt anatomical atlases to the patient's images, accounting for tumor growth and concomitant healthy tissue deformation. We also propose to demonstrate the utility of this methodology in building statistical atlases that link position and shape of the tumor, surgical approach, and surgical outcome, by applying it to a small group of patients with tumors near motor and speech areas and correlating surgical outcome with tumor position and geometry. The proposed methodology combines a biomechanical model of the brain with a statistical model that captures systematic relationships between a patient's anatomy prior to tumor growth and the patient's anatomy after tumor growth, thereby mathematically parameterizing the underlying brain deformation. The ability to utilize mathematical models of brain anatomy and to select from a group of standard approaches according to an objective prediction of the amount of brain injury each entails, will significantly improve clinical outcome of common neurosurgical procedures.